All skills

Reads a US 401k plan packet and converts fund fees, match fine print, and vesting schedules into concrete dollar figures.

Repairs hard-wrapped Claude Code .txt exports by rebuilding tables, paragraphs, paths and tool output.

Turns any UI or design into a structured WCAG 2.2 audit with pass/fail checks and a prioritised remediation plan.

A director-grade skill that builds launch films, explainers and motion graphics as editable HTML + GSAP code and renders them to MP4/GIF.

Turns textbook PDFs, stories and ideas into narrated SVG animated web books, documentaries, vertical shorts and explorable maps.

Restructures a bloated CLAUDE.md/AGENTS.md into a lean Level 1 plus on-demand Level 2 references with guaranteed zero information loss.

Moves Claude Code's per-project memory into tool-agnostic reference docs so Codex (and other AI CLIs) can read your profile, then verifies it by actually running codex.

Cleans up speech-to-text transcripts through a two-stage pipeline — instant dictionary rules plus Claude's own AI correction — while building a personal correction database that gets smarter with every fix.

Searches Claude Code (and Codex) session history across every config home and registered archive, and recovers deleted files byte-for-byte from file-history snapshots.

Generates stills, films, 3D motion graphics and web animations entirely in code across 32 styles, complete with composed scores, narration and subtitles.

Adversarially tears down a founder, company or product you envy — separating marketing bubble from real signal, then mapping the validated playbook onto your own resources.

Builds stills, animations and 3D motion graphics entirely in code across 32 styles, with composed scores, narration and subtitles.

Find, control, and screenshot specific macOS application windows programmatically using Swift, AppleScript, and screencapture.

Build, test and drive iOS apps in the Simulator with 31 token-efficient scripts that navigate by accessibility label instead of pixel coordinates.

Turns real logged agent failures into trace-based behavioural test cases plus a one-command Python runner.

Reads computed styles in the running app to catch hardcoded colors, off-palette backgrounds, unclickable controls, and animations that never ran.

When the console is empty and the code looks fine but the app still doesn't work, this skill gathers evidence from the running page and hands you the file:line to open.

Installs a dev-only SDK into your running web app so the agent can read real DOM, network, console and framework state, drive one real user flow end to end, and return a pass/fail verdict with the file:line to fix.

Sweeps a running web app by clicking every reachable control and reports dead buttons, console errors, failed requests and API/UI contradictions — no codebase knowledge required.

Measures real AI-engine referral traffic in GA4 to show which pages ChatGPT, Perplexity, Gemini and Claude actually cite — and what to fix next.